The subtle difference between "aid" and "finance" was lost in the launch of the database (one can hardly blame the media for this, as AidData began their report on the database by asking "How much aid does China give to Africa?" Their own name suggests that they are focusing on official aid, not commercial loans and export credits...).
I also learned that the figure of $75 billion was quoted by another Chinese official at an event at Brookings this week. (h/t to Winslow Robertson). So it will go as this rubbery figure takes on a life of its own (despite any corrections to the data produced by cloud-sourcing). And corrections are badly needed. By my estimates, even including all types of official government-to-government finance from China, the total should be well under $50 billion for the period covered by AidData. The errors are almost certainly not evenly spread across the dataset, which renders it rubbish for cross-country regressions.
If the Chinese published their aid data to Africa, we would have somewhere to go for the official figures. Yet who would listen? The Chinese actually did publish their global aid figures in 2011. They said that over the course of decades, China had committed approximately $37.7 billion in official development aid globally. Likely, about half of this would have gone to Africa. But this figure is now lost in the hoopla surrounding the $75 billion. Why isn't anyone asking the question: is it plausible that the Chinese government would have committed $37.7 billion in aid globally by the end of 2009, and $75 billion just to Africa, by the end of 2011?